Upjohn Institute Working Papers

Upjohn Research home page

2001

The Role of Temporary Agency Employment in Tight Labor Markets Susan N. Houseman W.E. Upjohn Institute, [email protected]

Arne L. Kalleberg University of North Carolina at Chapel Hill

George A. Erickcek W.E. Upjohn Institute, [email protected]

Upjohn Institute Working Paper No. 01-73 **Published Version** Industrial and Labor Relations Review 57 (October 2003): 105-127

Citation Houseman, Susan N., Arne L. Kalleberg, and George A. Erickcek. 2001. "The Role of Temporary Agency Employment in Tight Labor Markets." Upjohn Institute Working Paper No. 01-73. Kalamazoo, MI: W.E. Upjohn Institute for Employment Research. http://research.upjohn.org/up_workingpapers/73

This title is brought to you by the Upjohn Institute. For more information, please contact [email protected].

The Role of Temporary Agency Employment in Tight Labor Markets Upjohn Institute Staff Working Paper 01-73

by Susan N. Houseman W.E. Upjohn Institute for Employment Research Arne L. Kalleberg University of North Carolina–Chapel Hill George A. Erickcek W.E. Upjohn Institute for Employment Research

December 2000 last revision, January 2003 JEL Classification Codes: J49, J21, J31

This research was supported with a grant from the Russell Sage Foundation. We wish to thank Lillian Vesic-Petrovic and Peter Einaudi for outstanding research assistance and Claire Black for her transcription of all interviews. David Autor, Peter Cappelli, and two anonymous referees provided particularly helpful suggestions on earlier drafts of this paper.

The Role of Temporary Agency Employment in Tight Labor Markets

Abstract This paper examines the reasons why employers used and even increased their use of temporary help agencies during the tight labor markets of the 1990s. Based on case study evidence from the hospital and auto supply industries, we evaluate various hypotheses for this phenomenon. In high-skilled occupations, our results are consistent with the view that employers paid substantially more to agency help to avoid raising wages for their regular workers and to fill vacancies while they recruited workers for permanent positions. In low-skilled occupations, our evidence suggests that temporary help agencies facilitated the use of more “risky” workers by lowering their wages and benefits and the costs associated with turnover. The use of agency temporaries in both high- and low-skilled occupations reduced the pressure on companies to raise wages for existing employees, and thereby may have contributed to the stagnant wage growth and low unemployment observed in the 1990s.

Temporary agency employment expanded steadily during the 1990s. According to Bureau of Labor Statistics establishment data, employment in help supply services, which primarily comprises temporary help agencies, increased from 1.2 percent of paid employment in 1990 to 2.6 percent in 2000. The share in temporary agency employment grew throughout this period of strong economic expansion and very low unemployment despite the fact that the majority of temporary agency workers express a desire for regular, permanent jobs (Cohany 1998). The preference for regular employment among most agency temporaries has led many to conclude that the growth of the share in temporary employment in the 1990s was most likely driven by employer demand. In this paper, we provide insights into why employers heavily rely on temporary agency workers in tight labor markets based on evidence from two industries: hospitals and auto supply manufacturers. The fact that the share in temporary agency employment moves procyclically has long been observed. The most common explanation for this pattern is that companies use agency temporaries to handle demand variability and to buffer core workers during downturns (Mangum, Mayall, and Nelson 1985; Abraham 1988; Kandel and Pearson 2001). Because agency temporaries are the first to be laid off in a recession, we would expect the share in temporary employment to rise during expansions and fall during recessions. Nevertheless, it is hard to understand why, in the aggregate, the share in temporary employment would continue to grow strongly after almost a decade of expansion if firms were simply using temporaries to buffer core workers against an anticipated downturn. Indeed, evidence from employer surveys, (Abraham 1988, 1990; Houseman 2001b) suggests the importance of other factors, including difficulty in finding qualified workers on their own and the desire to screen workers for permanent positions. The hypothesis that, in tight labor markets, companies are filling vacancies with temporary

agency workers while they recruit employees and are sometimes recruiting employees from the ranks of agency temporaries is strongly supported by our case study evidence. One theme developed in this paper is that the use of agency temporaries in tight labor markets relieved pressure on companies to raise wages of regular employees. During the 1990s, the U.S. economy experienced low wage growth coupled with low unemployment rates.1 Although temporary agency employment still accounts for a relatively small share of paid employment, the growth in temporary employment accounted for 10 percent of net employment growth in the economy during the 1990s. If workers face costs in switching employers and internal labor market rules characterize wage setting for workers with some tenure on the job, then market forces primarily influence wage levels by affecting the wages of new entrants.2 Thus, it is reasonable to hypothesize some connection between the rapid growth in temporary agency employment, slow aggregate wage growth, and low unemployment. Indeed, Katz and Krueger (1999) present evidence that states with a greater presence of temporary agency employment experienced lower wage growth in the 1990s and that the lower wage growth associated with the growth in temporary employment may have accounted for up to a 0.4-percentage-point reduction in the unemployment rate. A major contribution of this paper is to improve our understanding of the mechanisms by which temporary agency use may lead to lower overall wage growth.

1

See Katz and Krueger (1999) for a documentation of this phenomenon and some hypotheses about its

causes. 2

See, for example, discussions on wage setting processes in internal labor markets in Doeringer and Piore 1971 and Levine, et al. 2002, Chapter 2.

2

WHY EMPLOYERS USE AGENCY TEMPORARIES IN TIGHT LABOR MARKETS: A CONCEPTUAL FRAMEWORK We begin by providing a conceptual framework for organizing and interpreting the case study information presented in subsequent sections. We assume that employers use agency temporaries in lieu of other factor inputs, such as direct-hire temporaries or regular employees, to reduce costs. Such cost savings can occur through a variety of mechanisms and we offer several potential explanations for why companies used and, on average, increased their use of temporary agency workers in the very tight labor markets characterizing the U.S. economy in the 1990s. One potential explanation is that companies often find it cost-effective to hire workers on explicitly temporary contract when the work is expected to be temporary or is of uncertain duration and companies have more such work when the economy is booming than when it is in recession. Theoretically, there must be some difference between temporary and regular workers to explain employers’ use of explicitly temporary contracts. To generate greater employer demand for temporary workers when an increase in workload is temporary or of uncertain duration, previous theoretical models (Abraham 1988, Kandel and Pearson 2001) have assumed that workers on temporary contract are less productive than regular workers but less costly to terminate.3 The amount of temporary work may increase during an expansion, as companies take on special projects that are intrinsically shortterm in nature. Moreover, during an expansion employers are more likely to experience an increase in 3

These termination costs can take various forms. U.S. employees have some rights to advance notice in the event of a mass layoff, and many state courts have granted workers legal protection against unjust dismissal in specific circumstances (Autor 2003). Moreover, companies may adopt personnel policies providing implied commitments of job security to regular employees in order to reduce turnover and build firm-specific human capital among a core set of workers. Independent of any legal restrictions, breaking these implied commitments may have costs, including lower morale and productivity and higher turnover. The assumption that regular employees are more productive–or have some other cost advantage–is needed to generate an economy with both regular and temporary workers.

3

their workload that, at least with some probability, is perceived as temporary. For instance, as employment increases during an expansionary period, employers may increase their relative use of temporary workers to buffer core workers, who are more expensive to lay off, against a possible downturn. This first explanation for why companies are more likely to use temporary workers when the economy is booming is akin to a scale effect: the greater quantity of temporary work will result in greater employer demand for all temporary workers: agency temporaries as well as temporary workers hired directly by companies.4 A second explanation is related to the fact that many companies use temporary agency workers to fill vacancies until permanent hires are made, in some cases, recruiting permanent workers from the ranks of the agency temporaries. The empirical importance of using agency temporaries to fill vacancies and to screen workers for permanent positions has been well established in recent surveys (Abraham 1990, Houseman 2001b, Kalleberg, Reynolds, and Marsden, 2003). Because agency temporaries are used for new hires and the accession rate moves procyclically, the share in temporary agency employment would also move procyclically. It is interesting to note that under this scenarios, the share in temporary employment would increase during an expansion even if the nature of the increased workload is not temporary. Finally, companies may substitute agency temporaries for direct-hire temporaries or for directhire recruits during expansions if the cost of agency temporaries relative to that of direct-hire temporaries or recruits declines when labor markets tighten. Here, the share in temporary employment 4

Abraham (1988) and Kandel and Pearson (2001) distinguish between temporary and permanent or regular workers, but not between types of temporary workers. The distinction between workers hired through third parties–like temporary agencies–and direct-hire temporaries is a critical component of the theory and empirical evidence presented below.

4

would grow during an expansion not only because the accession rate was increasing but also because the share of new hires who were agency temporaries was increasing. Regarding the last two explanations, there are various reasons why temporary agencies may enjoy a cost advantage in the recruitment and screening of temporary and permanent workers and why their cost advantage may increase when labor markets tighten. When labor markets are slack, it is relatively easy for companies to form in-house, on-call worker pools or to hire temporaries directly. When labor markets tighten, the supply of workers willing to accept temporary work or unpredictable hours shrinks at any given wage. Because temporary agencies pool jobs across companies, they can offer workers more steady employment, or for those who desire temporary work, agencies are better able to offer them employment when they want to work. Similarly, temporary agencies likely enjoy scale economies in recruiting and screening workers for permanent positions, especially in tight labor markets when companies have fewer unsolicited, qualified job applicants and receive fewer qualified applicants for any recruiting expenditure. In addition, we argue that the use of temporary agency workers to fill temporary positions or to fill vacancies while companies recruit permanent workers is often advantageous to companies because they may discriminate between temporary agency and direct-hire workers in the compensation offered. However, the direction of the wage discrimination along with the precise mechanisms by which temporary agencies are able to recruit workers and companies are able to save on compensation costs will depend on the labor market.5 We outline two scenarios below.

5

We use the term “wage discrimination” to denote the payment of unequal wages or compensation to agency temporary and permanent workers who are assigned to the same job.

5

•

Agency temporaries receive higher wages than direct-hire employees. Suppose, for now, that the workers an employer hires for a particular position are

homogeneous in quality. The market wage rate, w, facing individual employers depends upon the degree of labor market tightness, proxied by the unemployment rate, µ,

δw < 0. δµ

We assume that an employer’s existing workforce faces some costs in switching jobs.6 Therefore, across some range, a decline in the unemployment rate will not force an employer to raise wages to prevent existing workers from quitting to take new jobs. Similarly, we assume that wage cuts cause serious morale and productivity problems and consequently an employer will not lower wages for existing employees when unemployment increases, at least across some range. With the standard assumption that workers within the firm receive the same wage (i.e., no wage discrimination), the marginal cost of a new hire at time t is simply the wage paid to existing workers when the unemployment rate has been stable or rising since the last hire. However, if a firm wishes to hire additional labor, L, and labor markets have tightened since it last hired, the marginal cost is the higher

 δw   δµ   Lt , wt +     δµ   δt 

(1)

δµ < 0. δt

wage it must offer the new worker, plus the pay raise it must offer existing workers: A cost-minimizing employer would prefer to wage-discriminate, offering higher wages only to new workers. With wage discrimination, the marginal cost of hiring additional labor is simply the wage 6

These adjustment costs could include information costs in finding out changes in wage rates or preferences for the status quo. In models with heterogeneous labor, they could include firm-specific human capital for which the worker receives some additional compensation.

6

of the new hire. It is generally assumed that employers do not practice wage discrimination, even if existing workers face high costs of switching jobs, because such a two-tiered wage system would be seen as unfair by workers and result in serious morale and productivity problems. However, we argue any adverse morale and productivity effects are likely mitigated when new workers, who make more money than existing workers, are hired through a third party, such as a temporary help agency. These adverse morale and productivity effects might be lower if regular employees are less knowledgeable about temporary agency workers’ wage levels than they are about other employees’ wage levels. It may be more costly for employees to gain information about wage levels in other companies–in this case temporary agencies–than in their own company, particularly if the referring agency is not local. Additionally, employees may have fewer social contacts with agency workers than with other employees, and hence opportunities to exchange wage information, particularly if agency workers are on short-term assignment. Alternatively, because the client company does not directly set the wages of the temporary workers, employees may not fully blame their employers for the wage differentials. Formally, suppose that the cost of hiring an additional worker and paying her more than existing workers is K if the worker is hired directly, but only k, k < K, if the worker is hired through a third party. Then, if the firm chooses to practice wage discrimination, it will only do so through a third party, like a temporary agency, and it will only practice wage discrimination in new hiring if the additional

 δw   δµ     L , exceed the marginal cost of wage discrimination, k.  δµ   δt 

payments to existing workers, 

7

Thus, when unemployment is falling, and hence wages of new hires are rising, some employers may choose to practice wage discrimination through a temporary agency. 7 The use of temporary agency workers may also make it possible for employers to discriminate in the composition of the compensation package offered in ways that would be illegal or impractical if all workers were employees. For instance, nondiscrimination clauses in the Employee Retirement Income Security Act (ERISA) and the Internal Revenue Service tax code make it difficult for employers to offer different benefits packages to different groups of full-time employees.8 Temporary help agencies, which typically offer much lower benefits levels than client firms, likely attract some workers who place a low value on company-provided benefits. For instance, workers with health insurance coverage from another family member may place little or no value on health insurance benefits offered by their employer. Therefore, if the cash component of compensation in temporary jobs is higher than the cash component in regular jobs, it is possible that temporary agency workers place a higher value on the compensation from the temporary agency than the compensation they would earn in a regular employment position, even if the total cost of providing that compensation package is the same or less. In these cases, the use of temporary help agencies allows efficient sorting of workers by compensation preference and potentially benefits workers and employers alike. Thus far, the choice between regular and temporary workers has been posed as a one-time decision by employers seeking to increase their workforces. One might more realistically think of the 7

Bellemore (1998) develops a similar model of wage discrimination to study wage differentials between agency and regular nurses. Our model differs from his in that it does not require that employers have monopsony power in labor markets; rather, in our model, employers face rising wages over time if labor markets tighten. 8

ERISA and non-discrimination clauses in the IRS tax code are designed to ensure that highly compensated individuals are not the primary beneficiaries of fringe benefits that constitute in-kind, tax-free or tax-deferred income. See Houseman (2001a) for further discussion of these rules.

8

process as a dynamic one in which firms temporarily use higher-priced temporary labor in order to buy additional time to fill permanent vacancies. Specifically, in a model in which workers have imperfect information about the distribution of potential wage offers, a firm’s probability of filling a vacancy at any point in time would vary positively with the wage package and the unemployment rate.9 Employers may quickly recruit high-priced temporary agency labor to fill vacancies, buying time to recruit permanent workers at lower wages.10 Whether or not employers choose to utilize temporary agency workers for this purpose will depend on the price of temporary relative to permanent workers, the time it would take to recruit regular workers at lower wages, and the cost of leaving vacancies unfilled. In this dynamic context, the essential theoretical insight is the same as in the static model discussed above: it may be advantageous for employers to hire temporary workers in tight labor markets because they can discriminate in the wages they offer temporary and permanent workers and thereby avoid raising wages for existing workers. •

Agency temporaries may receive lower wages than direct-hire employees. Consider now a labor market in which there are two types of workers: L1 are “good” workers,

who have good work histories and are certain to work out; L2 are “risky” workers who have poor or no work histories, but, with probability ρ , will prove to be “good” workers, 0 < ρ < 1. Wage levels for good workers are greater than or equal to those of risky workers, w1 > w2. Risky workers who

9

The assumption that vacancies and thus employment levels are a function of the wage rate has been used in a number of theoretical models. For a review of these models, see Card and Krueger (1995). 10

Note that even if temporary workers are paid more, some workers may still prefer permanent positions. This is because temporary and permanent positions are not identical–the latter is associated with more job security and possibly more benefits. Thus, for some workers, the non-compensation aspects of a permanent job outweigh the lower compensation, or their personal valuation of the total compensation package may be higher for the permanent job.

9

prove themselves to be good workers receive w1 after a probationary period. Risky workers who prove to be “bad” workers quit or are fired, with a cost f to the employer; f captures any direct costs of dismissal as well as recruitment and screening costs associated with turnover. Expected output is assumed to be a simple linear function of the two potential labor inputs: E [Q ] = L1 + ρL2 When hiring, cost-minimizing employers will select a good or a risky worker depending upon the cost per unit of output for each type.

(2)

 w2 t + (1 − ρ ) f  w1t >  , < ρ

(3)

 w2t + (1 − ρ ) f   δw   δµ  w1t +  1    L1t >  , < δµ δ t ρ   

δµ ≥0 δt

δµ